RAT-Mania: Advancing Multi-RAT Networks With Optimal Beamforming, Slot, and Power Allocation, Through Optimization, Heuristics, and Learning

In an increasingly interconnected world, multiple technologies operating in the unlicensed band necessitate innovative resource allocation strategies for seamless coexistence. This paper proposes and rigorously analyzes a resource allocation framework for beamforming, independent of precoding matrices, enabling multiple radio access technologies and radio access points to coexist efficiently. The proposed framework is examined in diverse simulated transport scenarios including a bus, train, and aircraft, while also demonstrating its generalizability in offices and parks. This paper simplifies the mathematical description of such networks, with the goal to optimize the beam angle allocation for the radio access points over time, with the help of a Mixed-Integer Nonlinear Programming solver like Gurobi. We further compare the achievable performance for sum rate and fairness of the Gurobi solution with that of two meta-heuristics, Genetic Algorithm and Sparrow Search Algorithm. To promote practical deployment, we propose a learning-based implementation with Proximal Policy Optimization. In addition to sum rate and fairness, an in-depth analysis of the energy consumption indicates significant energy savings through additionally optimizing the transmission power of the radio access points.